Stacking of interferometric data - a submillimetre and radio view on the evolution of distant galaxies

Doctoral thesis, 2016

Understanding the processes of galaxy evolution requires observational constraints on the physical properties of galaxies at different times in the history of the universe. Large and deep surveys at visible and near-infrared wavelengths have produced extensive catalogues of high-redshift galaxies, spanning a large range of the history of the universe. Over this time the galaxies undergo significant evolution, increasing not only their stellar content, but also their physical size. In this thesis I will present results from observations of star-forming galaxies at submm and radio wavelengths. Observations at these longer wavelengths probe young stars, showing where new stars are formed in the galaxies. The observations presented in this thesis show that, for star-forming galaxies out to z ≈ 3, the sizes measured at submm and radio wavelengths are significantly smaller than those measured at near-infrared wavelengths. This implies that most stars are formed in the centre of galaxies, indicating that in the absence of other size evolution mechanisms we expect the typical effective radii of galaxies to decrease with time. It highlights the need for other size evolution mechanisms, such as minor mergers or changes in the galaxies due to stellar feedback. A major part of this thesis investigates the technique of stacking for interferometric data. Stacking is a technique to study statistical properties of populations, and is currently essential for the study of high-redshift, star-forming galaxies at submm and radio wavelengths, as many of high-redshift galaxies are too faint to be observed directly. Typically, stacking at different wavelengths ranges is done using deep imaging surveys observed with a single telescope. However, interferometry is not a direct imaging technique, and this presents a number of challenges to stacking. We present a new stacking algorithm that works directly on the visibilities; we refer to it as uv-stacking. We compare this algorithm to an image-stacking algorithm, i.e., an algorithm that stacks the sources directly in the imaged data. The uv-stacking algorithm is found to yield more robust results than the image-stacking algorithm. It is of particular interest for size measurement of stacked galaxies, as it preserves the uv data through stacking, and allows for robust model fitting of the stacked data.